I find exposing to the right and pulling down highlights works pretty well to be honest I just feel it is not as big a deal as it is being made out to be which causes a follow on effect of misinformation being repeated as fact
Ok, fair enough. Totally valid for you if the current system works for you! But it
can be game-changing for others. After all, a 3-stop difference is the difference between having to use a 3-stop Grad ND filter or not... or the difference between having to use, say, a 4stop reverse + a 3stop grad ND filter or just one of those in a really high dynamic range scene... in combination, these filters give you horrid magenta casts (even Singh-Ray ones... stacking just doesn't work well). You can also rescue more improperly exposed shots (bound to happen in fast-paced shooting scenarios), etc. etc. All been covered before.
5D Mark III's shadows (mine anyway) actually starts looking poor even earlier due to pattern noise (banding). briansquibb's 1ds3, however, has a remarkable absence of banding (from the dark frame he sent me), so he may have more salvageable files.
BTW, you may already know this, but, the idea that there's some magical extra DR in highlight headroom is a bit of a fallacy. Proper DR testing methodologies (like DXO, I'm assuming) account for this. I myself account for it in my tests by shooting multiple exposures 1/3EV apart, then finding the shot that is 1/3EV short of having, say, G1 green channel blown for the brightest wedge. That ensures that any variability in highlight headroom due to the manner in which the luminosities are mapped in the RAW file does
not impact the DR calculation.